Optical control of genetically defined protein activities, i.e. optogenetic regulation of proteins, has long been a dream in biology. If we could develop a generalizable method to optically control activities of proteins of interest, it could profoundly transform biological experimentation and impart new capabilities to gene- and cell-based therapies. For instance, the cellular functions that proteins coordinate, such as survival, apoptosis, differentiation, migration or connectivity (in the nervous system) could be controlled in vivo to study organismal physiology with micron-level resolution. Therapeutically implanted cells could be similarly controlled by light to precisely focus treatment at desired anatomical locations. There are countless other possible applications. Given these advantages of optical protein regulation, considerable efforts have been expended to adapt known light-responsive signaling domains to regulate mammalian proteins. However, existing strategies require extensive screening to create light-responsive proteins, or rely on protein relocalization to indirectly regulate activity. As a result, these methods have been used to control only a few proteins. Thus there exists a need for a method to create light-regulated proteins of interest that is generalizable. We have recently discovered a new class of light-mediated protein-protein interaction, and translated this discovery into a generalizable method for controlling protein activities with light. We hypothesized that fluorescent proteins (FPs) could undergo light-dependent conformational changes that drive changes in oligomerization state. Indeed, we found that tetrameric and dimeric variants of the reversibly photoswitching green FP Dronpa undergo dissociation as they are switched from bright to dark states by cyan light. We then discovered that fusing Dronpa domains at both ends of an enzymatic domain of interest cages it in the dark but allows uncaging upon illumina